Nvidia has obtained clearance from Beijing to sell its secondary artificial intelligence processors in China and is reportedly engineering a tailored version of the Groq AI chip for that market, according to Reuters.
This pivotal regulatory breakthrough enables the U.S. semiconductor leader to resume deliveries of its H200 series. These specific chips have served as a major point of contention in U.S.-China relations, representing a regional market that previously contributed 13% of Nvidia’s total revenue.
While Washington had authorized the exports and Chinese enterprises expressed strong interest, the primary hurdle to shipping H200 hardware has been Beijing’s internal hesitation to grant import permits. Nvidia intends to utilize Groq chips for inference tasks, such as answering queries, coding, or executing user commands. In recent product demonstrations, the firm revealed plans to pair its upcoming Vera Rubin chips — which are restricted from sale in China — with these Groq components.
Although Nvidia maintains a monopoly on AI system training, the inference sector presents steeper competition. Prominent Chinese entities, including AI leaders like Baidu, are already manufacturing proprietary inference hardware. On Tuesday, Nvidia CEO Jensen Huang announced that the H200 had been licensed for “many customers in China” and that the company had secured numerous purchase orders, facilitating a return to production for the chip.
“Our supply chain is getting fired up,” Huang said at a press conference.
The firm had suspended manufacturing of the processor last year due to escalating regulatory obstacles in both the United States and China. Nvidia had spent months awaiting licensing from both governments. Having secured various U.S. approvals, a source familiar with the situation confirmed that Beijing has now issued licenses for a broad range of Chinese clients.
CNBC corroborated on Tuesday that Huang confirmed the company holds clearance from both nations.
While a source at a Chinese firm indicated they were uncertain if Beijing had granted final official consent, they noted that Nvidia had invited them to begin placing purchase orders, reported Reuters.
In a late-month SEC filing, Nvidia disclosed that the U.S. government issued a license in February permitting “small quantities of H200 products to specific China-based customers.” Previously, Reuters reported in January that preliminary approval was granted to major tech players — including ByteDance, Tencent, and Alibaba — as well as startup DeepSeek, though the specific regulatory terms were still being finalized at that time.
Huang’s optimistic remarks regarding the AI agent OpenClaw, which has seen rapid integration in China, helped drive several Chinese AI-related stocks to record peaks on Wednesday. Valuation for large-language model startups MiniMax and Zhipu AI climbed over 19% following Huang’s assertion that OpenClaw represents “definitely the next ChatGPT.”
